# Project Manager: AI That Runs Your Development Projects > The Project Manager is an AI project manager for software development in the Ai1 platform by MyZone AI: it breaks your build requests into steps, routes each to your developer agent with full context, gets every change tested, reviews what comes back and shows you live progress; a live deploy always waits for your go-ahead. Canonical page: https://myzone.ai/pages/agents/ai-project-manager-agent Part of Ai1, by MyZone AI. Book a Project Manager walkthrough: https://calendly.com/d/ct6h-tcy-8qf/ai1-demo?a1=Project%20Manager&utm_source=myzone.ai&utm_medium=agent-page&utm_content=ai-project-manager-agent-final Last updated: 2026-10-05. Reviewed by the MyZone AI team. Describe the change. It plans, routes, tests and keeps you in the loop. ## At a glance - **Planning:** Splits each request into steps and works out what can run together - **Routing:** Sends each piece to your developer agent with pages, screenshots and instructions - **Testing:** A tester agent checks every change; failures go straight back - **Progress:** A live view of what is in progress, done and queued - **Where:** In chat or by posting in Slack - **What it never does:** Write code itself, or deploy live without your go-ahead ## What the Project Manager does - **Breaks requests into clear steps:** Splits what you want built or fixed into separate pieces and works out which can run together and which must wait. - **Routes work with full context:** Sends each piece to your developer agent with the project, the relevant pages, reference screenshots and the right instructions. - **Reviews what comes back:** Checks the developer's summary and sends work back with details if it needs another pass. You do not have to manage that loop. - **Gets every change tested:** Hands finished work to a tester agent to check behaviour and capture screenshots. Failures go straight back to the developer. - **Shows you where things stand:** Gives you a live view of what is in progress, what is done and what is queued. - **Commits and deploys when you say so:** Handles the commit once work is confirmed, and always asks your go-ahead before anything goes live. - **Works with your Notion:** Reads your Notion wiki and trackers, and adds notes or updates rows only after you say yes. ## How the Project Manager differs from a task board A task board records what needs doing and waits for someone to move the cards. The Project Manager moves the work itself: it plans each request, sends it to your developer agent with full context, gets it tested, sends failures back, commits the result and shows you live progress. Independent tasks run in parallel, dependent ones wait, and nothing reaches production without your go-ahead. ## How it works 1. **You make a request** (Any time): Describe what you want built, changed or fixed, in chat or by posting in Slack. 2. **Plan and assign** (Right away): It breaks the work down and sends each piece to your developer agent with full context. 3. **Build and test** (For every change): The developer agent builds the change, and a tester agent checks it against the expected behaviour. 4. **Review** (Before anything is committed): It reviews the results and sends anything that failed back for another pass, then re-tests. 5. **Commit and deploy** (you approve) (When you say so): It commits the finished work and deploys to staging when asked. A live deploy always waits for your approval. ## When to use it - **A bug needs fixing:** The developer agent fixes it, the tester confirms it works and it commits the change, with screenshots and a clear commit message for you. - **A feature touches several pages:** It runs the pages in parallel through the developer, verifies each one and commits once they all pass. - **You want to push staging to live:** It confirms the work is complete, asks for your go-ahead and then runs the live deploy. - **You post a request or brief in Slack:** It picks it up, confirms the plan with you, gets approval and routes the work to the right place. - **A test comes back with failures:** It sends the failure details to the developer automatically and re-tests once the fix is in. ## What you get - A step-by-step plan for each request - Work routed to your developer agent with full context - Tested changes with screenshots - Descriptive commit messages for every change - A live view of progress across your project - Staging and live deploys, with live always approved by you ## Example: release project plan and status Example with a fictional company. Names, people and figures are invented to show the agent's output. Any resemblance to a real company or person is unintended. What it was asked: Is our Mobile 3.0 release on track for 12 November? Give me the milestone plan, where this sprint stands, the risks, and what I need to decide this week. For a real client the plan is built from the team's task board, sprint history, budget tracker and meeting notes, and each status is checked against the work itself rather than taken on trust. This example uses an invented company, so no real systems or data were read and every name, date and figure is made up; the forecasts are recomputed from those invented figures and cross-checked, and no industry facts are cited. Run date: 29 September 2026. ### What it found - The Mobile 3.0 release is at risk: 60% of the work is done and spending is on plan, but at today's pace the last day for new features slips a week, to 6 November. - The slip has two causes, not effort: a payments task blocked for three days by a supplier's test system, and one code reviewer, so finished work waits two days. - Recommendation: move saved cards to the January update to keep the 12 November launch at US$409,000, under the US$440,000 budget; moving the date instead leaves one working day before the peak-season freeze. ## Guardrails - It never writes code or edits project files itself. All changes go through your developer agent. - Every change is tested before it is committed, and failures go back for fixing. - Live deploys always need your explicit go-ahead before anything reaches production. - Commits run through one point of control, so you always know what changed and why. ## Frequently asked questions ### What is an AI project manager for software development? An AI project manager coordinates build work instead of doing it. The Project Manager in Ai1 by MyZone AI breaks requests into steps, routes them to your developer agent, gets each change tested, reviews what comes back and commits the result. Live deploys wait for your approval. It does not replace your task board; it runs the work on it. To start, it needs your project set up on the Ai1 platform, a developer agent assigned to it and a tester agent for the checks. ### How does AI coordinate developer and tester agents? The Project Manager sends each piece of work to your developer agent with the project, the relevant pages, screenshots and instructions. A tester agent then checks it against the expected behaviour. Anything that fails goes back to the developer automatically with the failure details and is re-tested once the fix is in. You do not need to manage that loop. ### Can I send the Project Manager requests through Slack, and do I approve every step? Yes. Mention the Slack app in your project channel, or attach a requirements brief, and it picks up the request, confirms the plan with you and routes the work. You do not approve every step, but live deploys always need your explicit go-ahead before anything goes to production. ### Can the Project Manager handle several changes at once? Yes. If changes touch different parts of the project, it runs them in parallel. If they share files or depend on each other, it puts them in the right order. Either way it never edits project files, HTML, CSS, JavaScript or config itself. All file changes go through your developer agent. ## More on this topic - [Everything I Learned About Agentic Development on a 6-Week Hyper-Focused Bender](https://myzone.ai/pages/blog/agentic-development-bender): Our founder went from zero coding to enterprise-level software in 6 weeks using AI agents. Here's everything he learned about agentic development. ## About Ai1 Ai1 is the AI operations platform by MyZone AI, where each client runs on its own private server. The Project Manager is included on every Ai1 level, including Developer Core, with no per-agent charge. It works alongside the other Ai1 agents on your account. Pricing: https://myzone.ai/pages/services/ai1-pricing. Security: https://myzone.ai/pages/security.